Rumor Detection over Varying Time Windows

By Sejeong Kwon, Meeyoung Cha, Kyomin Jung. PLoS ONE, 2017

345 citations recorded by the source corpus.

This study determines the major difference between rumors and non-rumors and explores rumor classification performance levels over varying time windows—from the first three days to nearly two months. A comprehensive set of user, structural, linguistic, and temporal features was examined and their relative strength was compared from near-complete date of Twitter. Our contribution is at providing deep insight into the cumulative spreading patterns of rumors over time as well as at tracking the precise changes in predictive powers across rumor features. Statistical analysis finds that structural and temporal features distinguish rumors from non-rumors over a long-term window, yet they are not available during the initial propagation phase. In contrast, user and linguistic features are readily available and act as a good indicator during the initial propagation phase. Based on these findings, we suggest a new rumor classification algorithm that achieves competitive accuracy over both short and long time windows. These findings provide new insights for explaining rumor mechanism theories and for identifying features of early rumor detection.

Subfields: Sociology and Political Science, Statistical and Nonlinear Physics

View Paper: https://www.semanticscholar.org/paper/7c0b1f319e2e432afeb48cfdb9ffd5102e97f119

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Sociology and Political Science

In Statistical and Nonlinear Physics

Citations in the corpus

A survey on fake news and rumour detection techniques (2019, background): “Other studies have focused instead on modeling the temporal characteristics of propagation [56, 57].”
Detection and Resolution of Rumours in Social Media (2017, methodology): “More recently, [Kwon et al. 2017] analyse feature stability over time and report that structural and temporal features distinguish true from false rumours over a long-term window.”
A Survey of Deep Active Learning (2020, background): “[22] KDnugget’s Fake News 8, Harvard Dataverse [124], Liar [234] News veracity detection”
Detect Rumors in Microblog Posts Using Propagation Structure via Kernel Learning (2017, methodology): “…inspired by prior analyses on rumors spreading, e.g., user information can be a strong clue in the initial broadcast, content features are important throughout entire propagation periods, and structural and temporal patterns help for longitudinal diffusion (Zubiaga et al., 2016; Kwon et al., 2017).”
Combating Fake News (2019, background): “Variants of the above include other network-based features that are slightly extended or tailored to a particular context, such as the inclusion of temporal features [56] or geographic location and type of device (mobile or PC) from which the response was sent [122].”
Fake News Identification on Twitter with Hybrid CNN and RNN Models (2018, neutral): “Rumors have been examined in the context of varying time windows [26].”
CSI: A Hybrid Deep Model for Fake News Detection (2017, background): “Branching o of the core linguistic analysis, many have combined the approach with traditional classiers to label an article as true or false [6, 11, 18, 25, 27, 41, 45].”
GCAN: Graph-aware Co-Attention Networks for Explainable Fake News Detection on Social Media (2020, background): “(4) RFC (Kwon et al., 2017): an extended random forest model combining features from retweet user proﬁles and the source tweet.”